Mailing List Archive

(no subject)
Hoi,

This is a request that has been waiting for an answer now for a month.
Personally I am of the opinion that we should positively answer this
request. Sander Spek is one of the most respected Dutch wikipedians, he
is the one mentoring Frank Schepers. If there is a need for extra surity
for the privacy of the table, we can ask for some acknowledgment that it
will be used only and exclusively for the use of this research.

Thanks,
GerardM
-------------------------------------------------
Frank Schepers
Keelstraat 5a
3770 Vroenhoven – Riemst (Belgium)
0032 – 498 – 365280

Hello,

I'm a last year student Knowledge Engineering/Computer Science of the
faculty of General
Sciences at the University of Maastricht. The subject of the Master
Thesis I'm working
on is: "Personalization of the WikiWikiWeb". On the Dutch Wikipedia I
have an account with
the nickname frankschepers, but I don't have edited articles yet. My
supervisor is prof.
dr. Jaap van den Herik. My daily advisor will be drs. Sander Spek, who
is a frequent editor
on the Dutch Wikipedia site and who came up with the subject of this
Master thesis. The
thesis will contain a theoretical and a practical part. In the
theoretical part various
methods of personalization will be examined and their applicability on
Wiki's.
In the practical part, I will make an extension for the wikimedia
software. This will be
an article recommender system. This system will display for a
Wikipedian, who has a watch
list, some articles that aren't on his watch list, but though are
interesting for him.
If it is possible to create a good working recommender system, this will
improve the quality
of the articles on Wikipedia because more interested users will
read/edit the article.

So I will evaluate if personalization can be applied in an effective
way on Wiki's.
For this research I need a database dump of the watchlist table. This
table cannot be downloaded
from http://download.wikimedia.org/archives/nl/ because of some privacy
issues I suppose.

So my question is, if it is possible to get a database dump of the watch
list table.

If you require any further information, feel free to contact me.

Kind regards,

Frank
Re: (no subject) [ In reply to ]
On Apr 9, 2005 10:35 AM, GerardM <gerard.meijssen@gmail.com> wrote:
>
> Hoi,
>
> This is a request that has been waiting for an answer now for a month.
> Personally I am of the opinion that we should positively answer this
> request. Sander Spek is one of the most respected Dutch wikipedians, he
> is the one mentoring Frank Schepers. If there is a need for extra surity
> for the privacy of the table, we can ask for some acknowledgment that it
> will be used only and exclusively for the use of this research.


Is there any reason he can't just come up with a random watchlist for made
up users? I guess the table could be given out with scrambled usernames, but
I wouldn't like the watchlist tables to be given out otherwise.
Re: (no subject) [ In reply to ]
> Is there any reason he can't just come up with a random watchlist for made
> up users? I guess the table could be given out with scrambled usernames, but
> I wouldn't like the watchlist tables to be given out otherwise.

Scrambled user names would not solve the privacy issue. It would still
be very easy to guess which watchlist belonged to which person based
on which articles they were watching. I doubt, for example, that
anyone else has all my user talk archives and other sub pages on their
watchlist other than me. Similiarly, most people watch the articles
they have created, so if only one person is watching [[Ajaria]],
[[Ajman]], and [[Batumi]], you can guess that this is probably Danny's
watchlist.

I think the only solution to this is to ask people to opt-in to the
study, and then the watchlists of only those people could be
extracted.

Angela.
Re: (no subject) [ In reply to ]
Hello

>Scrambled user names would not solve the privacy issue. It would still
>be very easy to guess which watchlist belonged to which person based
>on which articles they were watching. I doubt, for example, that
>anyone else has all my user talk archives and other sub pages on their
>watchlist other than me.
>
Ermmmm on the Dutch wikipedia I have everybody's talkpage on my
watchlist. Keeps me on track on discussions. But I am probably the
exception .... :(

>
>I think the only solution to this is to ask people to opt-in to the
>study, and then the watchlists of only those people could be
>extracted.
>
>
He can have my watchlist data it has over 4000 watched pages, so let;s
see if that helps him further.

Walter/Waerth
Re: (no subject) [ In reply to ]
On 4/15/05, Walter van Kalken <walter@vankalken.net> wrote:

> He can have my watchlist data it has over 4000 watched pages, so let;s
> see if that helps him further.

If his plans are what I think they are, he will need watchlists from
quite a large number of people, and having large watchlists from a few
people will not be a replacement. On the other hand, I do think that
there would be good possibilities not working with watchlists at all.

If his plans are not what I think they are, I'd better consider this
message not sent...

Andre Engels